Download Animal Carcinogenicity Studies: 2. Obstacles to Extrapolation of

The Humane Society Institute for Science and Policy
Animal Studies Repository
2-2006
Animal Carcinogenicity Studies: 2. Obstacles to
Extrapolation of Data to Humans
Andrew Knight
Animal Consultants International, [email protected]
Jarrod Bailey
University of Newcastle-upon-Tyne
Jonathan Balcombe
Physicians Committee for Responsible Medicine, [email protected]
Follow this and additional works at: http://animalstudiesrepository.org/acwp_arte
Part of the Animal Experimentation and Research Commons, Animal Studies Commons, and the
Other Medical Sciences Commons
Recommended Citation
Knight, A., Bailey, J., & Balcombe, J. (2006). Animal carcinogenicity studies: 2. Obstacles to Extrapolation of Data to Humans. ATLANOTTINGHAM-, 34(1), 29.
This Article is brought to you for free and open access by the Humane Society Institute for Science and Policy. It has been accepted for inclusion by an
authorized administrator of the Animal Studies Repository. For more information, please contact [email protected].
ATLA 34, 29–38, 2006
29
Animal Carcinogenicity Studies: 2. Obstacles to
Extrapolation of Data to Humans
Andrew Knight,1 Jarrod Bailey2 and Jonathan Balcombe3
Consultants International, London, UK; 2School of Population and Health Sciences, Faculty of
Medical Sciences, University of Newcastle upon Tyne, Newcastle upon Tyne, UK; 3Physicians Committee for
Responsible Medicine, Washington DC, USA
1Animal
Summary — Due to limited human exposure data, risk classification and the consequent regulation of
exposure to potential carcinogens has conventionally relied mainly upon animal tests. However, several
investigations have revealed animal carcinogenicity data to be lacking in human predictivity. To investigate
the reasons for this, we surveyed 160 chemicals possessing animal but not human exposure data within
the US Environmental Protection Agency chemicals database, but which had received human carcinogenicity assessments by 1 January 2004. We discovered the use of a wide variety of species, with rodents
predominating, and of a wide variety of routes of administration, and that there were effects on a particularly wide variety of organ systems. The likely causes of the poor human predictivity of rodent carcinogenicity bioassays include: 1) the profound discordance of bioassay results between rodent species, strains
and genders, and further, between rodents and human beings; 2) the variable, yet substantial, stresses
caused by handling and restraint, and the stressful routes of administration common to carcinogenicity
bioassays, and their effects on hormonal regulation, immune status and predisposition to carcinogenesis;
3) differences in rates of absorption and transport mechanisms between test routes of administration and
other important human routes of exposure; 4) the considerable variability of organ systems in response to
carcinogenic insults, both between and within species; and 5) the predisposition of chronic high dose
bioassays toward false positive results, due to the overwhelming of physiological defences, and the unnatural elevation of cell division rates during ad libitum feeding studies. Such factors render profoundly difficult any attempts to accurately extrapolate human carcinogenic hazards from animal data.
Key words: animal experiment, animal test, bioassay, cancer prevention, carcinogenicity, chemical
classification, chemical safety, extrapolation, risk assessment.
Address for correspondence: A. Knight, Animal Consultants International, 91 Vanbrugh Court, Wincott
Street, London SE11 4NR, UK.
E-mail: [email protected]
Introduction
Due to a paucity of human exposure data, public
health policy and the regulation of exposures to
potential human carcinogens by governmental
agencies such as the US Environmental Protection
Agency (EPA) have for decades relied mainly upon
animal carcinogenicity tests. However, our survey
of 160 environmental contaminants considered to
be of greatest concern within the USA, listed in the
EPA’s Integrated Risk Information System (IRIS)
chemicals database, that lacked significant human
exposure data but possessed animal data, found
that the EPA considered the animal data inadequate to support a classification of probable human
carcinogen or non-carcinogen in the majority
(58.1%; 93/160) of cases (1). Significant differences
in human carcinogenicity classifications of identical
chemicals between the EPA and the World Health
Organisation’s International Agency for Research
on Cancer (IARC) indicate that the true predictivity
for human carcinogenicity of animal data is even
less than that indicated by EPA figures alone.
The sensitivity of the conventional rodent bioassay
to human carcinogens (the ability to detect them) for
some sex–species combinations is not in question.
However, it’s very poor human specificity (the ability
to identify human non-carcinogens) has also been
revealed by other investigators (2–6), and contributes to the poor human predictivity of the assay,
that is, the low probability that a positive result is
truly indicative of human carcinogenicity. To investigate the reasons for the poor human specificity and
predictivity of animal carcinogenicity tests, we examined the IRIS chemicals that had received EPA
human carcinogenicity assessments primarily on the
basis of animal carcinogenicity data.
Methods
Of the 543 chemicals contained in the IRIS chemicals database as of 1 January 2004, 160 lacked significant human exposure data, but possessed
animal data and had received human carcinogenicity assessments. For each of these 160 chemicals,
A. Knight et al.
30
we determined the species and routes of administration used, and the organ systems affected.
Results
Number of routes
of administration
Species used
The species used were not provided for two of the
160 chemicals, both of which were considered
unclassifiable as to their human carcinogenicity.
The other 158 chemicals were each studied in up to
seven different species (in the case of N-nitrosodiethylamine), with the most common numbers of
species used per chemical (Table 1) being two
(67.7%), one (14.6%, many, but not all of which
resulted in a human carcinogenicity assessment of
“unclassifiable”), and three (11.4%).
At least 10 different species were used for these
158 chemicals, namely: chickens, dogs, guinea-pigs,
hamsters, mice, mink, primates (one macaque,
three unspecified “monkey” species, and one
unspecified “primate” species), rabbits, rats and
trout. The three species most commonly used per
chemical (Figure 1) were mice (92.4%), rats (86.7%)
and hamsters (14.6%).
Routes of administration
Routes of administration were not provided for four
of the 160 chemicals. Up to 10 routes of administration (in the case of benzo[a]pyrene) were used for
each of the other 156 chemicals, with the most common numbers of routes used per chemical (Table 2)
being one (43.6%), two (21.1%) and three (19.2%).
Table 1: Species used with assessed EPA
chemicals lacking significant
human data but possessing animal
data
Number of
species used
Table 2: Routes of administration for
assessed EPA chemicals lacking
significant human data but
possessing animal data
Number of
chemicals
% of total
chemicals
1
2
3
4
5
6
7
23
107
18
6
0
3
1
14.6
67.7
11.4
3.8
0.0
1.9
0.6
Total
158
Data source: The EPA Integrated Risk Information
System database, 1 January 2004.
1
2
3
4
5
6
10
Total
Number of
chemicals
% of total
chemicals
68
33
30
16
4
4
1
43.6
21.1
19.2
10.3
2.6
2.6
0.6
156
Data source: The EPA Integrated Risk Information
System database, 1 January 2004.
Twelve non-oral routes of administration and a
variety of oral routes, not always specified, were
used (Figure 2), namely: dermal, inhalation, intramuscular, intraperitoneal, intrapleural, intrarenal,
intratesticular, intravenous, oral (food), oral (gavage), oral (water), oral (other, e.g. capsule, toothpaste additive), oral (unspecified), subcutaneous,
surgical implantation, transplacental, and vaginal
painting. Those most commonly used were food
(49.4%), gavage (33.3%) and dermal administration
Table 3: Organ systems affected by
chemicals assessed by the EPA to
be probable (B2) or possible (C)
human carcinogens
Number of organ
systems affected
Number of
chemicals
% of total
chemicals
1
2
3
4
5
32
21
17
9
5
30.8
20.2
16.3
8.7
4.8
6
7
8
9
10
11
9
5
2
1
1
2
8.7
4.8
1.9
1.0
1.0
1.9
Total
104
Data source: The EPA Integrated Risk Information
System database, 1 January 2004.
Obstacles to extrapolation of animal carcinogenicity data to humans
31
Figure 1: Species used with assessed EPA chemicals lacking significant human data but
possessing animal data
100
92.4
86.7
90
% of total chemicals (158)
80
70
60
50
40
30
20
14.6
8.2
10
0
mouse
rat
hamster
dog
4.4
3.8
3.2
rabbit
guineapig
primate
0.6
0.6
0.6
chicken
mink
trout
species
Data source: The EPA Integrated Risk Information System database, 1 January 2004.
(26.3%). Other routes of major interest were drinking water (21.1%) and inhalation (17.9%).
Organs affected
Of the 160 chemicals, those considered probably not
carcinogenic to humans were not, of course, known
to exhibit significantly neoplastic lesions. For the
56 chemicals considered unclassifiable, it was frequently difficult to establish whether or not significant treatment-related results occurred. However,
for the remaining 104 chemicals, considered probable or possible human carcinogens, up to 43 organs
or organ systems were found to exhibit neoplastic
lesions (Figure 3). However, given that only
selected organ systems were examined by some
investigators, and given that the location of several
gastrointestinal and respiratory tract neoplasms
were unspecified, the true prevalence of neoplastic
lesions may have been even higher.
Up to 11 organ systems (in the case of 1,2-dibromoethane and N-nitrosodiethylamine) were recorded
as exhibiting neoplastic lesions for each chemical
(Table 3), although most commonly one (30.8%), two
(20.2%) or three (16.3%) organ systems were affected.
For some chemicals, unclear reporting may have
resulted in duplication of results under multiple categories. For example, in a study of dietary aniline
hydrochloride administered to rats, the investigators
reported, among other tumours, statistically signifi-
cant dose-related increases of haemangiosarcomas in
the spleen, fibrosarcomas and sarcomas in the body
cavity and spleen, and fibrosarcomas and sarcomas in
multiple organs of the body cavity. The resulting classifications included connective tissue, peritoneal cavity, spleen and vascular system (7).
The organ systems most commonly affected
(Figure 3) were the liver (66.3%), the lung (31.7%),
and the kidney, skin and stomach (all 17.3%).
Differentiation between primary and metastatic
tumours was often impossible, hence all tumours
were included, even when (infrequently) identified as
metastatic.
Discussion
Species used
As stated, the 158 applicable EPA chemicals were
tested in up to seven species each, although two
(67.7%), one (14.6%) or three (11.4%) species were
most commonly used (Table 1). Although at least 10
different animal species were used for these chemicals (Figure 1), the most commonly-used animals
were mice (92.4%) and rats (86.7%).
Discordance between mice and rats
Intuitively, one might expect mice and rats to display similar carcinogen susceptibilities. However,
A. Knight et al.
32
Figure 2: Routes of administration for assessed EPA chemicals lacking significant human
data but possessing animal data
oral: food
49.4
oral: gavage
33.3
dermal
26.3
subcutaneous
22.4
intraperitoneal
21.1
oral: water
21.1
routes of administration
inhalation
17.9
oral: unspecified
9.0
surgical implantation
5.8
transplacental
3.2
intramuscular
2.6
oral: other
2.6
intravenous
1.9
intrapleural
0.6
intrarenal
0.6
intratesticular
0.6
vaginal painting
0.6
0
10
20
30
40
50
60
% of total chemicals (156)
Data source: The EPA Integrated Risk Information System database, 1 January 2004.
while they do, for some chemicals, for others, they
do not. For example, Hengstler et al. (8) described
the striking variation in aflatoxin B1 carcinogenesis
susceptibility between rats and mice, with mice
refractory to dietary levels three orders of magnitude higher than those found to be carcinogenic in
rats. Murine resistance to aflatoxin B1 results from
an efficient conjugation with glutathione, catalysed
by glutathione S-transferase mYc, a biochemical
mechanism apparently lacking in the rat.
In 2001, Gottman et al. (9) compared 121 replicated rodent carcinogenicity bioassays conducted
Obstacles to extrapolation of animal carcinogenicity data to humans
33
Figure 3: Organ systems affected by chemicals assessed by the EPA to be probable (B2) or
possible (C) human carcinogens
66.3
liver
31.7
lung
kidney
17.3
skin
17.3
stomach
17.3
16.3
thyroid gland
15.4
mammary gland
10.6
haematopoietic system
9.6
adrenal gland
9.6
nasal cavity
8.7
testes
injection site
7.7
urinary bladder
7.7
6.7
lymphatic system
6.7
vascular system
5.8
biliary system
4.8
organ systems affected
oesophagus
connective tissue
3.8
peritoneal cavity
3.8
spleen
3.8
thorax
3.8
trachea
3.8
3.8
uterus
ear/zymbal’s gland
2.9
large intestine
2.9
larynx
2.9
nervous system
2.9
oral cavity
2.9
pituitary gland
2.9
pancreas
1.9
preputial gland
1.9
respiratory tract, unspecified
1.9
salivary gland
1.9
small intestine
1.9
subcutaneous tissue
1.9
gastrointestinal tract, unspecified
1.0
jaw, lower
1.0
mesothelioma
1.0
nasopharynx
1.0
ovary
1.0
parathyroid gland
1.0
pharynx
1.0
thymus
1.0
0
10
20
30
40
% of total chemicals (104)
Data source: The EPA Integrated Risk Information System database, 1 January 2004.
50
60
70
A. Knight et al.
34
both within the US National Cancer Institute/
National Toxicology Program (NCI/NTP) and elsewhere, and published and recorded within the comprehensive Berkeley-based Carcinogenic Potency
Database (CPDB). They found a reproducibility of
only 57% overall for rodent carcinogenicity classifications between the NCI/NTP and studies from the
general biomedical literature. This value did not
improve substantially when additional biological
information (species, strain, sex, target organs) was
considered. Ettlin and Prentice (10), Fung et al.
(11) and Johnson (12) have also demonstrated that
bioassays involving strains other than those used by
the NTP give discordant results.
Haseman (2) examined the long-term rodent
studies of 385 substances recorded within the
NCI/NTP database. After eliminating equivocal
results, 207 chemicals were found to be carcinogenic in at least one sex–species group, yet only 56
of these (27.1%) were carcinogenic in both male and
female mice and rats. Male rats appeared to be
more susceptible to tumourigenesis than females,
but the situation was reversed in mice. Di Carlo
(13) similarly found that of 61 chemicals causing
cancer in mice or rats, only 13 (21.3%) were carcinogenic in both male and female mice and rats. Di
Carlo concluded that “it is painfully clear that carcinogenesis in the mouse cannot now be predicted
from positive data obtained from the rat and vice
versa.”
Discordance between rodents and primates
The high carcinogenesis predisposition of rodents
when compared to primates (3, 14) further complicates the extrapolation of results to humans.
Detailed analyses of 25 rodent carcinogens tested in
monkeys for up to 32 years by the NCI, revealed
that half were not monkey carcinogens, in most
cases despite strong evidence of carcinogenicity in
rodents and/or humans (15). Another 26-year NCI
study of model rodent carcinogens found that only
urethane was carcinogenic in monkeys (16).
Numerous important differences between
rodents and humans impact on predisposition to
carcinogenesis. These differences between rats and
humans include mean lifespan (2.5 versus 70
years), food consumption (50 versus 10g/kg/day),
basal metabolic rate (109 versus 26kcal/kg/day),
anatomical differences (the forestomach, Zymbal’s
gland, Harderian gland, preputial gland and clitoral
gland exist only in the rat), stomach pH (4–5 versus
1–2), and, very significantly, DNA excision repair
rates (low versus high) (17).
Species differences in absorption, distribution,
metabolism and elimination pathways or rates can
all influence the carcinogenicity of a chemical. Since
many carcinogens must be metabolised to reactive
electrophiles to produce their carcinogenic effects,
species differences in pathways or rates can affect
carcinogenic metabolite activity levels (18).
Examples of rodent carcinogens considered unclassifiable as to human carcinogenicity by the IARC,
because their mechanisms of carcinogenesis are not
present in humans (19–21), include d-limonene
(rodent mechanism — renal tubular alpha2u-globulin mediated nephrotoxicity), saccharin and its
salts, and melamine (rodent mechanism — urolithiasis predisposed by rodent urinary composition).
Indeed, it is remarkable that mice can develop
very malignant tumours with multiple genetic
alterations within 6–18 months, whereas aggressive
tumours in humans or other primates may take
many years to reach an equivalently life-threatening stage (22). Some 50% of all chemicals tested for
carcinogenicity in mice or rats are positive in at
least one experiment, with predisposition to carcinogenesis even higher in some commonly-used
strains (3, 23). Holliday (24) suggested that the high
predisposition of rodents to carcinogenesis when
compared to humans, might be due to less-efficient
DNA repair, poorer control of genetic stability,
and/or altered control of gene expression. The high
doses applied in rodent bioassays may also increase
apparent carcinogenicity. The metabolic pathways
and rates of activation and detoxification may
become saturated at high tissue concentrations,
resulting in differences in target tissue doses,
altered tissue responses, and variations in the
organs affected (18).
Of all of the species used, primates are the most
biologically similar to humans. However, their lack
of availability, high cost of procurement, housing
and maintenance, and their lengthy life cycles, all
serve to minimise their use (16). In contrast, the
widespread availability, low costs of procurement,
housing and maintenance, and the short life cycles
of rodents, thereby facilitating lifetime exposure
and developmental deformity studies, result in
their predominant use in carcinogenicity bioassays.
However, the largely logistical benefits of rodents
do not qualify them scientifically as the test species
of choice. In fact, the profound discordance of bioassay results between rodent species, strains and genders, and further, between rodents and human
beings, means that it is profoundly difficult to make
human carcinogenicity assessments on the basis of
rodent bioassay data.
Routes of administration
Carcinogenesis predisposition of stressful routes
Studies on mice, rats, hamsters, monkeys, dogs, rabbits, birds, and even bats, have shown that basic nonor minimally-invasive procedures, such as handling
or gavaging (administration of a test substance via
Obstacles to extrapolation of animal carcinogenicity data to humans
35
an oesophageal tube), cause significant increases in
stress indicators, including concentrations of glucose, corticosterone (a stress hormone), growth hormone, noradrenaline, prolactin, thyroid-stimulating
hormone and triiodothyronine (25). Other blood
measures, including packed cell volume, haemoglobin and plasma protein, also rise significantly (26).
These stress-related responses generally occur with
every exposure to such a stressor, and laboratory
animals do not readily habituate to them (25).
Stress-related responses are particularly important in long-term carcinogenicity studies, in light of
their frequent use of stressful routes of administration. Gavaging was used for 33.3% of the 156 applicable EPA chemicals under consideration, and
dermal administration (requiring handling and
restraint) for 26.3% (Figure 2). Other routes of
administration used, which required handling and
restraint as a minimum, were intramuscular,
intraperitoneal, intrapleural, intrarenal, intratesticular, intravenous, oral (other than food or water,
for example, via capsule or toothpaste additive),
subcutaneous, surgical implantation and vaginal
painting.
The stress-mediated hormonal changes and disruption of normal hormonal regulation that occur in
response to such stressful stimuli, predispose animals
to immunosuppression and increased susceptibility
to virtually all pathologies, including neoplasia.
Moynihan et al. (27) documented immunosuppression following handling, as evidenced by decreased
immunoglobulin G levels in mice. Brenner et al. (28)
documented more pulmonary metastases in 11 similarly-handled female mice than in non-handled controls, following intravenous challenge with tumour
cells.
Tumour size can also be affected unpredictably
by stressors, which complicates any assessments.
Fifteen female mice handled daily (picked up by the
tail and held gently in the palm for two minutes) on
days one to five of tumour growth, had increases in
tumour size compared to non-handled controls,
while 13 mice similarly-handled on days one to
seven did not (29).
Intra-laboratory and inter-laboratory variations
in environmental conditions are also likely to result
in varying hormonal disruptions and consequent
immunosuppression. For example, animals are
positioned in varying proximity to stressful stimuli
such as doors, telephones, air-conditioning ducts
and light bulbs. For some years, NTP rodent bioassay protocols used different shelves in the animal
house for different treatment groups, potentially
skewing the results (30).
To standardise and eliminate such variations in
environmental conditions would be a very largescale and potentially expensive undertaking.
Furthermore, it would not eliminate the variable
but substantial stresses caused by handling and
restraint, and the stressful routes of administration
endemic to carcinogenicity bioassays, nor their
inevitable effects on hormonal regulation, immune
status, predisposition to carcinogenesis, and bioassay results.
Route-to-route extrapolation
Judgements frequently need to be made about the
carcinogenicity of a chemical via a route of exposure
different to that studied. For example, exposures of
interest may sometimes be through the inhalation
of a chemical tested primarily through feeding studies (31). Given that only 17.9% of the chemicals of
interest were tested via inhalation, in contrast to
the percentages tested via food (49.4%), gavaging
(33.3%) or drinking water (21.1%), such dilemmas
hardly seem unlikely.
Quantitative extrapolation between routes of
exposure is frequently problematic. Differences in
rates of absorption and transport mechanisms
between routes (for example, oral, inhalation, dermal) can be great. There is no generally applicable
method for accounting for these differences in
uptake processes, although confidence is strengthened when the tumour effects are observed at a site
distant from the point of entry, or when absorption
through the route of interest is similar to that via
the tested route (31).
Organs affected
The 104 EPA chemicals considered to be probable
or possible human carcinogens induced neoplasia in
43 different organ systems (Figure 3), with those
most commonly affected being the liver (66.3%), the
lung (31.7%), and the kidney, skin and stomach (all
17.3%). Up to 11 different organ systems were
affected by each chemical, although most commonly
one (30.8%), two (20.2%) or three (16.3%) organ
systems were affected (Table 3).
Interspecies variation
The wide variation in organ systems affected may
have been exacerbated by the considerable variability of organ systems and species in response to
many carcinogenic insults. Comparisons between
mice, rats, hamsters and humans, for example,
reveal that carcinogens are carcinogenic at the
same site in another of these species no more than
50% of the time (3). Patterns of tumour incidence,
whether spontaneous or in response to carcinogenic
insults, differ dramatically between rodents and
humans, and even between different rodent strains.
High incidences of rare human tumours, such as
those of the liver, pituitary and testis, occur spontaneously in various strains of mice or rats, whereas
A. Knight et al.
36
common human tumours, such as those of the
prostate and colorectum, occur rarely, if ever, in
rodents (32–33).
Dose-related toxicity
In an attempt to interpret differing carcinogenic
incidences across a large number of organ systems,
investigators commonly assume that human carcinogenic risk is proportional to the number of
organ systems affected, and that it increases when
multiple sex–species groups are affected or when
fatalities result. However these assumptions were
undermined by Lois et al. (34), who found, in their
analysis of around 4000 chronic carcinogenicity
bioassays on 1050 chemicals, that carcinogens
affecting multiple sex–species groups or causing
fatalities are likely to be the same chemicals that
affect multiple organ systems, and by Meijers et al.
(35), who found that neoplastic lesions in multiple
organ systems are more likely to be indicative of
dose-related toxicity than true carcinogenicity.
Despite critics such as Monro and Davies (36),
who point out that, since human carcinogens are
amongst the most potent of rodent carcinogens it
should be possible to detect them by using relatively
low dose levels, carcinogenicity bioassays typically
rely upon the maximum tolerated dose (MTD), as
indicated by increasing toxicity-related effects, in
order to maximise their sensitivity to carcinogens.
However, prolonged exposure to high chemical
doses can result in chronic irritation, cell death, and
consequent cellular proliferation (mitogenesis).
Sodium saccharin and related sodium and potassium salts, for example, produce urinary calculi,
and consequently, cellular proliferation, regenerative hyperplasia and neoplasia, only at high doses
(3, 37–38).
Animals have a broad range of physiological
defences against general toxic insults, such as
epithelial shedding and inducible enzymes, which
commonly prove effective at environmentally relevant doses, but which may be overwhelmed at
higher doses (3). When combined with insufficient
rest intervals between doses for the effective operation of DNA and tissue repair mechanisms, the outcome can be an increased predisposition to
mutagenesis and carcinogenesis. Lower doses,
greater intervals between exposures, or shorter
total periods of exposure, some or all of which represent a more realistic model of true environmental
exposures for most potential carcinogens, might not
result in carcinogenesis at all.
Calorie-induced mitogenesis
Reviews of both the experimental and epidemiological literature show a high correlation between
increased cell division and carcinogenesis. Ad libitum feeding, as occurs in many carcinogenicity
studies, can unnaturally elevate cell division (3),
thereby increasing predisposition to carcinogenesis,
with consequent increases in false positive rates.
Including indices of cell division for both test and
control groups would facilitate the identification of
groups with elevated cell division rates due to ad
libitum feeding, thereby allowing the subsequently
increased predisposition to carcinogenesis to be
accounted for during the interpretation of the
results. If necessary, feeding protocols could also be
altered. Unfortunately, however, indices of cell division are not normally included or considered (3).
False positive results
It is consequently unsurprising that, for rodent carcinogens tested at the MTD and half the MTD,
around half the organs in which tumours arise are
only affected at the MTD (3). Clearly, chronic highdose carcinogenicity bioassays, in which physiological defences are overwhelmed and ad libitum
feeding results in the unnatural elevation of cell
division rates, are inherently predisposed towards
false positive results. Such factors have been implicated as the cause of the poor human specificity of
animal carcinogenicity studies, as revealed by other
investigators (2–6) and ourselves (1).
Conclusions
Our survey of the 160 chemicals that had received
EPA human carcinogenicity assessments primarily
on the basis of their animal carcinogenicity data,
revealed that a wide variety of species, predominantly rodents, were used, with a wide variety of
routes of administration, and with effects on a particularly wide variety of organ systems.
The likely causes of the poor human specificity,
and hence low predictivity, of rodent carcinogenicity bioassays revealed by other investigators (2–6)
and ourselves (1), include: 1) the profound discordance of bioassay results between rodent species,
strains and genders, and further, between rodents
and humans; 2) the variable, yet substantial
stresses, caused by handling and restraint and the
stressful routes of administration common to carcinogenicity bioassays, with consequent effects on
hormonal regulation, immune status and predisposition to carcinogenesis; 3) the differences in transport mechanisms and rates of absorption between
test routes of administration and other important
human routes of exposure; 4) the considerable variability of organ systems affected in response to carcinogenic insults, between and within species; and
5) the inherent predisposition of chronic high-dose
bioassays toward false positive results, likely to be
Obstacles to extrapolation of animal carcinogenicity data to humans
37
due to the overwhelming of physiological defences,
and the unnatural elevation of cell division rates
during ad libitum feeding. The EPA has acknowledged some of the difficulties on its IRIS website:
Carcinogenic Risks to Humans. Volumes 1–55. Lyon,
France: International Agency for Research on
Cancer.
National Cancer Institute (1978). Bioassay for Aniline
Hydrochloride for Possible Carcinogenicity. CAS No.
142-04-1. ITS Carcinogenesis Technical Report Series
No. 130. Department of Health, Education and Welfare
(DHEW) Publication No. NIH 78-1385. 113pp. Bethesda, MD, USA: DHEW.
Hengstler, J.G., Van der Burg, B., Steinberg, P. &
Oesch, F. (1999). Interspecies differences in cancer
susceptibility and toxicity. Drug Metabolism Reviews
31, 917–970.
Gottmann, E., Kramer, S., Pfahringer, B. & Helma, C.
(2001). Data quality in predictive toxicology: reproducibility of rodent carcinogenicity experiments.
Environmental Health Perspectives 109, 509–514.
Ettlin, R.A. & Prentice, D.E. (2002). Unexpected
tumour findings in lifetime rodent bioassay studies
— what to do. Toxicology Letters 128, 17–33.
Fung, V.A., Barrett, J.C. & Huff, J. (1995). The carcinogenesis bioassay in perspective: application in
identifying human cancer hazards. Environmental
Health Perspectives 103, 680–683.
Johnson, F.M. (1999). Carcinogenic chemical-response
‘‘fingerprint’’ for male F344 rats exposed to a series of
195 chemicals: implications for predicting carcinogens
with transgenic models. Human Ecology and Risk
Assessment 5, 427–443.
Di Carlo, F.J. (1984). Carcinogenesis bioassay data:
correlation by species and sex. Drug Metabolism
Reviews 15, 409–413.
Goodman, J.I. (2001). A perspective on current and
future uses of alternative models for carcinogenicity
testing. Toxicologic Pathology 29, Suppl. 1, 173–176.
Gold, L.S., Manley, N.B., Slone, T.H. & Rohrbach, L.
(1999). Supplement to the Carcinogenic Potency
Database (CPDB): results of animal bioassays published in the general literature in 1993 to 1994 and
by the National Toxicology Program in 1995 to 1996.
Environmental Health Perspectives 107, Suppl. 4,
527–600.
Schoeffner, D.J. & Thorgeirsson, U.P. (2000).
Susceptibility of nonhuman primates to carcinogens
of human relevance. In Vivo 14, 149–156.
Monro, A. & Mordenti, J. (1995). Expression of exposure in negative carcinogenicity studies: dose/body
weight, dose/body surface area, or plasma concentrations? Toxicologic Pathology 23, 187–198.
Dybing, E. & Huitfeldt, H.S. (1992). Species differences in carcinogen metabolism and interspecies
extrapolation. In Mechanisms of Carcinogenesis in
Risk Identification (ed. H. Vainio, P.N. Magee, D.B.
McGreger & A.J. McMichael), pp. 501–522. Lyon,
France: International Agency for Research on Cancer.
International Agency for Research on Cancer (1999).
IARC Monographs Programme on the Evaluation of
Carcinogenic Risks to Humans. Vol. 73: Some Chemicals that Cause Tumours of the Kidney or Urinary
Bladder in Rodents, and Some Other Substances.
674pp. Lyon, France: International Agency for
Research on Cancer. Website http://monographs.
iarc.fr (Accessed 20.01.04).
Hard, G.C. & Whysner, J. (1994). Risk assessment of
d-limonene: an example of male rat-specific renal tumorigens. Critical Reviews in Toxicology 24, 231–254.
Environmental Protection Agency (1991). Report of
the EPA Peer Review Workshop on Alpha2u-globulin: Association with Renal Toxicity and Neoplasia
“In general IRIS values cannot be validly used to
accurately predict the incidence of human disease or
the type of effects that chemical exposures have on
humans. This is due to the numerous uncertainties
involved in risk assessment, including those associated with extrapolations from animal data to humans
and from high experimental doses to lower environmental exposures. The organs affected and the type of
adverse effect resulting from chemical exposure may
differ between study animals and humans.” (39).
Such factors result in the poor human specificity,
and hence low predictivity, of animal carcinogenicity data, and render profoundly difficult any
attempts to accurately extrapolate human carcinogenic hazards from animal data.
Fortunately, alternative carcinogenicity testing
strategies exist, which have the potential to offer
superior human specificity, as well as substantially
reduced time-frames and greatly reduced demands
on financial, personnel and animal resources. These
alternatives are explored in detail in Knight et al.
(40).
Acknowledgement
7.
8.
9.
10.
11.
12.
13.
14.
15.
This research was partly funded by the Physicians
Committee for Responsible Medicine, Washington
DC, USA.
Received 6.8.05; received in final form 18.12.05; accepted
for publication 19.12.05.
16.
17.
References
1.
2.
3.
4.
5.
6.
Knight, A., Bailey, J. & Balcombe, J. (2006). Animal
carcinogenicity tests: 1. poor human predictivity.
ATLA 34, 19–27.
Haseman, K. (2000). Using the NTP database to
assess the value of rodent carcinogenicity studies for
determining human cancer risk. Drug Metabolism
Reviews 32, 169–186.
Gold, L.S., Slone, T.H. & Ames, B.N. (1998). What
do animal cancer tests tell us about human cancer
risk? Overview of analyses of the carcinogenic
potency database. Drug Metabolism Reviews 30,
359–404.
Ashby, J. & Purchase, I.F. (1993). Will all chemicals
be carcinogenic to rodents when adequately evaluated? Carcinogenesis 8, 489–495.
Rall, D.P. (2000). Laboratory animal tests and
human cancer. Drug Metabolism Reviews 2,
119–128.
International Agency for Research on Cancer
(1972–1992). IARC Monographs on the Evaluation of
18.
19.
20.
21.
A. Knight et al.
38
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
in the Male Rat. 93pp. Washington DC, USA: Environmental Protection Agency.
Balmain, A. & Harris, C.C. (2000). Carcinogenesis in
mouse and human cells: parallels and paradoxes.
Carcinogenesis 21, 371–377.
Gold, L.S. & Zeiger, E., eds (2000). Handbook of
Carcinogenic Potency and Genotoxicity Databases.
770pp. Boca Raton, FL, USA: CRC Press.
Holliday, R. (1996). Neoplastic transformation: the
contrasting stability of human and mouse cells.
Cancer Surveys 28, 103–115.
Balcombe, J.P., Barnard, N. & Sandusky, C. (2004).
Laboratory routines cause animal stress. Contemporary Topics in Laboratory Animal Science 43, 42–51.
Gärtner, K., Büttner, D., Döhler, K., Friedel, R.,
Lindena, J. & Trautschold, I. (1980). Stress response
of rats to handling and experimental procedures.
Laboratory Animal 14, 267–274.
Moynihan, J., Brenner, G., Koota, D., Breneman, S.,
Cohen, N. & Ader, R. (1990). The effects of handling
on antibody production, mitogen responses, spleen
cell number, and lymphocyte subpopulations. Life
Sciences 46, 1937–1944.
Brenner, G.J., Cohen, N., Ader, R. & Moynihan, J.A.
(1990). Increased pulmonary metastases and natural
killer cell activity in mice following handling. Life
Sciences 47, 1813–1819.
Aarstad, H.J. & Seljelid, R. (1992). Effects of stress
on the growth of a fibrosarcoma in nu/nu and conventional mice. Scandinavian Journal of Immunology 35, 209–215.
Young, S.S. (1989). What is the proper experimental unit for long-term rodent studies? An examination of the NTP benzyl acetate study. Toxicology
54, 233–239.
Anon. (1999). Guidelines for Carcinogen Risk Assess-
32.
33.
34.
35.
36.
37.
38.
39.
40.
ment. 211pp. Washington DC, USA: Risk Assessment
Forum, Environmental Protection Agency.
Monro, A. (1993). How useful are chronic (life-span)
toxicology studies in rodents in identifying pharmaceuticals that pose a carcinogenic risk to humans?
Adverse Drug Reactions and Toxicological Reviews
12, 5–34.
Monro, A. (1993). The paradoxical lack of interspecies correlations between plasma concentrations
and chemical carcinogenicity. Regulatory Toxicology
and Pharmacology 18, 115–135.
Lois, S.W., Slone, T.H., Manley, N.B., Garfinkel, G.B.,
Hudes, E.S., Rohrbach, L. & Ames, B. (1991). The
Carcinogenic Potency Database: analyses of 4000
chronic animal cancer experiments published in the
general literature and by the U.S. National Cancer
Institute/National Toxicology Program. Environmental Health Perspectives 96, 11–15.
Meijers, J.M., Swaen, G.M. & Bloemen, L.J. (1997).
The predictive value of animal data in human cancer
risk assessment. Regulatory Toxicology and Pharmacology 25, 94–102.
Monro, A. & Davies, T.S. (1993). High dose levels are
not necessary in rodent studies to detect human carcinogens. Cancer Letters 75, 183–194.
Cohen, S.M. & Lawson, T.A. (1995). Rodent bladder
tumors do not always predict for humans. Cancer
Letters 93, 9–16.
Cohen, S.M. (1995). Human relevance of animal carcinogenicity studies. Regulatory Toxicology and
Pharmacology 21, 75–80.
Anon. (2006). IRIS limitations. Website www.epa.
gov/IRIS/limits.htm (Accessed 05.02.06)
Knight, A., Bailey, J. & Balcombe, J. (2006). Animal
carcinogenicity tests: 3. alternatives to the bioassay.
ATLA 34, 39–48.